J Appl Phycol (2014) 26:491–504 DOI 10.1007/s10811-013-0052-1
Distribution patterns and introduction pathways of the cosmopolitan brown alga Colpomenia peregrina using mt cox3 and atp6 sequences Kyung Min Lee & Ga Hun Boo & James A. Coyer & Wendy A. Nelson & Kathy Ann Miller & Sung Min Boo
Received: 31 January 2013 / Revised and accepted: 9 May 2013 / Published online: 24 May 2013 # Springer Science+Business Media Dordrecht 2013
Abstract Colpomenia peregrina is an annual brown macroalga found in temperate waters worldwide. To understand population differentiation and to reconstruct pathways of colonization/introduction, we analyzed variation in two mitochondrial protein-coding genes, cytochrome c oxidase subunit III (cox3) and ATP synthase F0 subunit 6 (atp6), and cp RuBisCO spacer. A total of 359 cox3, 342 atp6, and 38 RuBisCO spacer sequences from Colpomenia peregrina were obtained for samples collected at 28 sites from 12 countries. The combined cox3+atp6 sequences (1,231 bp) revealed 99 polymorphic sites and 69 haplotypes. An mt haplotype network revealed four distinct groups, separated by 7 to 26 mutation steps. NW Pacific populations were present in each group (but dominant in one), whereas SW Pacific and the Atlantic populations each were present in one group. The network and phylogenetic analyses, along with patterns of genetic diversity, suggested a NW Pacific center of origin, expanding first to the SW Pacific, then the Electronic supplementary material The online version of this article (doi:10.1007/s10811-013-0052-1) contains supplementary material, which is available to authorized users. K. M. Lee : G. H. Boo : S. M. Boo (*) Department of Biology, Chungnam National University, Daejeon 305-764, Korea e-mail:
[email protected] J. A. Coyer Shoals Marine Laboratory, Cornell University, 400 Little Harbor Road, Portsmouth, NH 03801, USA W. A. Nelson National Institute of Water and Atmospheric Research, Wellington 6241, New Zealand K. A. Miller University Herbarium, University of California at Berkeley, Berkeley, CA 94720, USA
NE Pacific, and most recently to the north Atlantic. A generalized skyline plot revealed a dramatic population expansion of the species ca. 20 kya. Keywords Colpomenia peregrina . Haplotype network . Marine invasion . Mitochondrial genes . Phaeophyceae . RuBisCO spacer . Phylogeography
Introduction Phylogeographic studies of benthic marine species with cosmopolitan or worldwide distributions can offer insights into ecological and evolutionary processes. At one end of the continuum, an apparently widespread species may actually comprise one or more “cryptic” species that can be detected only with molecular markers. At the other end of the continuum, the species may possess efficient and effective mechanisms for long-distance dispersal and colonization, thereby promoting extensive gene flow among widespread locations. Both endpoints may be reinforced by anthropogenic introductions: single introductions may lead to a genetic bottleneck, differentiation via genetic drift, and, ultimately, speciation, whereas repeated introductions may promote gene flow and admixture of genotypes worldwide. The focus of the present study is the brown macroalga Colpomenia peregrina Sauvageau (Scytosiphonaceae), a cosmopolitan species ranging from Korea, Japan, China, and Russia in the NW Pacific (Kogame and Yamagishi 1997; Cho et al. 2005; Kozhenkova 2009; Boo 2010) to Alaska, USA, and Baja California, Mexico (Pedroche et al. 2008; Lindeberg and Lindstrom 2010) in the NE Pacific and New Zealand and Australia in the SW Pacific (Clayton 1979; Parsons 1982). In the north Atlantic Ocean, C. peregrina occurs from Norway to Portugal (Minchin 1991) and
492
from Labrador/Newfoundland to the Gulf of Maine (Kennedy et al. 2010). The annual C. peregrina exhibits a biphasic, heteromorphic life history consisting of saccate (5–10 cm) gametophytes and crustose sporophytes (Clayton 1979; Kogame and Yamagishi 1997). Reproduction occurs sexually, as well as by parthenogenesis of asexual zoospores. Thus, long-distance dispersal likely occurs by floating thalli that release either gametes or asexual spores and/or by reproductive individuals that are epiphytic on other species of floating algae. Human activity, especially intercontinental shipping and oyster mariculture, has increased the frequency of species exchange among continents and oceans and thereby the probability of marine algae crossing ecogeographic barriers. A single or a few founder individuals may initiate macroalgal invasions, or the patterns may be more complex with invasive populations composed of multiple genetic lineages resulting from multiple founder events (Mineur et al. 2010). Pathways of introduction have been investigated for several species of macroalgae using molecular markers: Ascophyllum nodosum (Olsen et al. 2010), Caulerpa taxifolia (Stam et al. 2006), Codium fragile (Provan et al. 2005a), Fucus distichus (Coyer et al. 2011b), F. serratus (Hoarau et al. 2007), F. spiralis and F. vesiculosus (Coyer et al. 2011a), Gracilaria vermiculophylla (Kim et al. 2010), and Undaria pinnatifida (Voisin et al. 2005). Members of the Scytosiphonaceae have a long history of transport around the globe through human activities. For example, five genera are regarded as human-mediated introductions in New Zealand and elsewhere: Chnoospora, Colpomenia, Hydroclathrus, Rosenvingea, and Scytosiphon (Johnson and Dromgoole 1977; Parsons 1982; Nelson and Duffy 1991, Cho et al. 2007; Nelson and Wilcox 2010). Of the 12 Colpomenia species, Colpomenia bullosa, Colpomenia claytoniae, and Colpomenia peregrina are considered invasive (Farnham 1980; Kain (Jones) et al. 2010; Boo et al. 2011b) in areas other than the NW Pacific. However, Colpomenia expansa and Colpomenia tuberculata, which are widely distributed in the NE Pacific (Pedroche et al. 2008), were reported recently in Korea (Lee 2008). Despite their widespread distributions, phylogeographic studies of Colpomenia spp. have not been attempted. In view of the cosmopolitan distribution of C. peregrina and the potential for long-distance dispersal via natural or anthropogenic means, our goals were to examine phylogeography and to identify pathways of introductions. We used two mitochondrial markers, the commonly used cytochrome c oxidase subunit III (cox3) and the rarely used (in algae) ATP synthase F0 subunit 6 (atp6) genes, both singly and concatenated. A subset of individuals from each population was further analyzed with chloroplast RuBisCO spacer region to serve as an independent assessment of introduction pathways.
J Appl Phycol (2014) 26:491–504
Materials and methods Specimens of Colpomenia peregrina (Fig. 1a–d) were obtained from 28 locations throughout the range of the species: NW Pacific (10 from Korea, two from Japan, and one from Russia), NE Pacific (one from Mexico and three from western USA), SW Pacific (two each from Australia and New Zealand), and the north Atlantic (one each from France, Ireland, Norway, Spain, and UK, and two from eastern USA). Thalli were collected at intervals of 1 m along a horizontal intertidal line for a total of 359 individuals from all populations. All collected individuals were stored in silica gel and later cleaned of epiphytes under a dissecting microscope in the laboratory. Voucher specimens were deposited in the herbarium of Chungnam National University, Daejeon, Korea and the herbarium of the Museum of New Zealand Te Papa Tongarewa, Wellington, New Zealand.
DNA extraction, amplification, and sequence alignment A total of 359 individuals were analyzed for mitochondrial cox3 and 342 for atp6. We also selected 38 individuals (at least one from each of the 28 locations) for plastid RuBisCO spacer. DNA was extracted from pulverized thalli using NucleoSpin® Plant II (Macherey-Nagel GmbH & Co, Germany) according to the manufacturer’s instructions. For amplification of cox3 gene, we used primers F49 and R20 (Boo et al. 2010, 2011a). Because atp6 (ATP synthase F0 subunit 6 gene) has not been used in marine algae, we designed primers F25P (5′-CCH TTA GAA CAA TTT BAA ATA CTY CC-3′) and R754P (5′-GCR TCR TTT ATR TAR ATR CAA CTT A-3′) based on published mitochondrial genome data from various brown algae: NC007684 for Desmarestia viridis, NC007685 for Dictyota dichotoma, NC007683 for F. vesiculosus, NC004024 for Laminaria digitata, NC003055 for Pylaiella littoralis, and NC93476 for Saccharina japonica. Thus, our atp6 primer set may be useful for other species of brown algae. The RuBisCO spacer region was amplified using primers RS1 and RS2 (Yoon and Boo 1999). PCR amplification, purification, and cycle sequencing were performed following Boo et al. (2011b). The sequences of forward and reverse strands were determined commercially (Genotech, Korea). Private haplotypes (haplotypes unique to a single population represented by a single mutation) were amplified twice and sequenced from a subset of samples to exclude the possibility of PCR errors. All gene sequences were aligned manually using SeAl v.2.0a11 (Rambaut 2002). Sequences were deposited in the NCBI GenBank with accession numbers JX027338-75 for cox3, JX027298-337 for atp6, and JX843461-70 for RuBisCO spacer.
J Appl Phycol (2014) 26:491–504
Phylogenetic analyses Phylogenies of cox3, atp6, the combined cox3+atp6, and RuBisCO spacer datasets were reconstructed using maximum likelihood (ML). MODELTEST 3.7 (Posada and Crandall 1998) indicated different models for different datasets: TrN+G for cox3, GTR+G for atp6 and mt combined cox3+atp6, and TrN+I for RuBisCO spacer. Because GTR is the most common and general model for realworld DNA analysis (Stamatakis 2006), we also used the GTR+G+I nucleotide model as implemented in RAxML v.7.0.4. However, because topologies of trees reconstructed by different models are similar to those of the GTR model, we employed the GTR model instead of different models for each gene. We used 200 independent tree inferences with the “number of run” option with default optimized SPR rearrangement and 25 distinct rate categories to identify the best tree. Statistical support for each branch was obtained from 1,000 bootstrap replications using the same substitution model and RAxML program settings. Maximum-parsimony (MP) trees were constructed for each data set with PAUP* v.4.0b.10 (Swofford 2002) using a heuristic search algorithm with the following settings: 1,000 random sequence additions, tree bisectionreconnection (TBR) branch swapping, MulTrees, all characters unordered and unweighted, and branches with a maximum length of zero collapsed. Bootstrap values for the resulting nodes were assessed using 1,000 bootstrapping replicates with 10 random sequence additions, TBR, and MulTrees. Trees were visualized using the FigTree v.1.1.2 program, available at http://tree.bio.ed.ac.uk/software/ figtree/. Eight Colpomenia species were used as outgroups in the mitochondrial analysis and two for the plastid gene, Fig. 1 Representative specimens of Colpomenia peregrina collected in a Sanjokam, Goseong, Korea (12 January 2005), b Ile de Batz, Roscoff, France (13 June 2010), c Santa Catalina Island, California, USA (17 July 2011), and d Appledore Island, Maine, USA (2 August 2011). Scale bars are 1 cm
493
based on their sister–group relationship to Colpomenia peregrina.
Population genetic analyses Haplotype (Hd) and nucleotide (π) diversities were measured for each population using DNASP V.5 (Librado and Rozas 2009). Relationships and geographic distributions of haplotypes were analyzed in a network constructed using TCS V.1.21 (Clement et al. 2000), which implements the statistical parsimony procedure with a 95 % connection limit. The hierarchical distribution of genetic variation among populations was tested using analysis of molecular variance (AMOVA) based on the number of pairwise nucleotide differences of the two mitochondrial loci (RuBisCO was not analyzed because of the much smaller dataset) using ARLEQUIN V.3.5 (Excoffier and Lischer 2010). We eliminated populations with less than ten samples to make population sizes more homogeneous. The null hypothesis of neutral evolution of the mt DNA was tested using Tajima’s D (Tajima 1989) and Fu’s Fs test (Fu 1997) with the program ARLEQUIN V.3.5. Significant D values can be due to factors such as selection, population expansion, and bottlenecking (Tajima 1989). Historical demographic expansions were investigated by examining the frequency distributions of pairwise differences between concatenated (cox3+atp6) sequences (mismatch distribution; Rogers and Harpending 1992). Mismatch distributions are used to test hypotheses on population demographic history and selection (Rogers and Harpending 1992). The distribution is usually multimodal in samples drawn from populations at demographic
494
equilibrium, but unimodal in populations following recent population demographic expansion and population range expansion (Slatkin and Hudson 1991). Historical demographic changes of NW Pacific populations, which is the origin center of C. peregrina (based on cox3 sequences because of availability of mutation rate data in the present study), were inferred using a generalized skyline plot (Strimmer and Pybus 2001), since the shape of a genealogy depends on the demographic history. The first step of this analysis is to generate a phylogenetic tree with branch lengths proportional to time. An ML tree using TrN+G model was estimated in PAUP v.4.0b10 (Swofford 2002). Next, a generalized skyline plot was generated from the ML tree using GENIE v.3.0 (Pybus and Rambaut 2002) with a smoothing algorithm to reduce the noise in the data while simultaneously preserving the demographic signal. The smoothing parameter (ɛ) was estimated using the “maximize optimization” option.
J Appl Phycol (2014) 26:491–504
differences between C. phaeodactyla and C. tuberculata) with 11.4 % average divergence. A total of 10 haplotypes of RuBisCO spacer (549 bp) were generated from 38 specimens (Table 1). Consequently, fewer positions were variable (14.8 %) and parsimoniously informative (4.4 %), with up to 1.0 % pairwise divergence within C. peregrina. The pairwise divergence ranged from 8.2 % (43 bp difference between C. peregrina and C. sinuosa) to 10.4 % (55 bp difference between C. peregrina and C. bullosa) with 9.6 % average divergence. Genetic diversity Genetic diversity was considerably lower in the north Atlantic than in all other areas (Table 2). For example, haplotype diversity (Hd) of cox3+atp6 sequences ranged from 0.92 in southern Korea to 0.00 in Spain and the eastern USA, whereas nucleotide diversity varied from 0.85 (northern Japan) to 0.00 (Spain and the eastern USA). Overall haplotype diversity was 0.93±0.07, and nucleotide diversity (π) was 0.86±0.03.
Results Haplotype network and divergence Characteristics of mtDNA and cp RuBisCO spacer sequences A total of 359 cox3 sequences were obtained (Table 1) for a 620-bp alignment with 204 variable positions (32.9 %) and 129 parsimoniously informative positions (20.8 %). Sequences of C. peregrina differed by up to 3.7 %, while the average sequence divergence within the genus was 11.2 %, ranging from 4.4 % (27 bp difference between Colpomenia peregrina and Colpomenia claytoniae) to 17.9 % (111 bp difference between Colpomenia phaeodactyla and Colpomenia tuberculata). A total of 342 atp6 sequences (Table 1) were generated for a 611-bp alignment with 200 variable positions (34.4 %) and 122 parsimoniously informative positions (21.0 %). Within C. peregrina, the sequences differed by up to 2.4 % pairwise divergence and within the genus Colpomenia by 4.8 %, ranging from 28 bp differences between C. peregrina and C. claytoniae to 18.6 % (108 bp differences between Colpomenia durvillei and C. tuberculata) with 11.9 % average divergence. Concatenation of cox3 and atp6 (1,231 bp) resulted in 69 haplotypes with 404 variable positions (32.8 %) and 251 positions (20.4 %) parsimoniously informative. Of the 69 haplotypes, 53 were privates, represented by a mutation of a single sequence and unique to a single population. Within C. peregrina, the sequences differed by up to 2.9 % pairwise divergence and within the genus Colpomenia by 4.6 % (55 bp differences between C. peregrina and C. claytoniae) to 18.2 % (218 bp
The concatenated haplotype network revealed four groups (Figs. 2 and 3) identical to the ML analyses of haplotypes (Fig. 4). Group 1 formed a star-like network with the core haplotype (h1) shared by 56 individuals among nine populations in the NW Pacific and one population in the NE Pacific (see Fig. S1 for haplotype networks of cox3 and atp6 singly). Group 1 haplotypes (h1 to h42) occurred in the NW and NE Pacific: haplotypes h27 to h35 were private to the NE Pacific (Santa Catalina Islands, USA and Baja California, Mexico), with h27 as the core. Group 2 consisted of haplotypes h43 to h46 found in the NW Pacific (southern Korea and northern Japan) and the NE Pacific (Alaska and Oregon, USA). Nine haplotypes (h47 to h55) formed group 3 and were present in the NW Pacific (Korea, northern Japan, and Russia) and the North Atlantic. Group 4 consisted of 14 haplotypes (h56 to h69) from Australia, New Zealand, and western Korea in a star-like network with h56 at the center. Haplotype networks for cox3 and atp6 were similar to one another, and each revealed the same four groups identified with the combined analysis (Fig. S1). A twogroup pattern was revealed within the RuBisCO spacer network (Fig. S1). Groups 2 (NW and NE Pacific) and 4 (NW and SW Pacific) were resolved with the RuBisCO spacer network, but not groups 1 and 3, undoubtedly because an order of magnitude fewer individuals were used for the analysis. Significant genetic concordance with biogeographic regions was found in C. peregrina (AMOVA, Table 3), and high
Collection location
SonoMX
Subtotal NE Pacific KruzUS SunsUS CataUS
OtagNZ
WellNZ
MckeAU
ShimJP HakoJP VladRU Subtotal SW Pacific BateAU
39
Kruzof Island, AK, USA Sunset Bay, OR, USA Santa Catalina Island, CA, USA Sonora, Mexico
Otakou, Otago harbour, New Zealand
3
3 1 24
34
10
11
7
6
Haengwonri, Jeju, Korea
HaenKR
13
Batemans Bay, New South Wales, Australia Mckenzie Bay, Sydney, Australia Wellington, New Zealand
Jeongdori, Wando, Korea
JeonKR
24
1 17 2 199
Dolsando, Yeosu, Korea
DolsKR
32
21
5 6 5 4
30
n cox3
Shimane, Honshu, Japan Hakodate, Hokkaido, Japan Vladivostok, Russia
Sangjokam, Goseong, Korea
SangKR
GyeoKR
Chongjin3ri, Pohang, Korea Suryeomri, Gyeongju, Korea Euihangri, Taean, Korea Daecheon harbor, Boryeong, Korea Gyeokpo, Buan, Korea
ChonKR SuryKR EuihKR DaecKR
NW Pacific AninKR Anin, Gangneung, Korea
Sample
c13(3) c13(1) c1(3), c15(16), c16(1), c17(2), c18(1), c20(1) c15(2), c19(1)
c28(8), c36(1), c37(1), c38(1) c28(6), c33(4)
c28(4), c32(1), c34(1), c35(1)
c28(3), c29(1), c30(1), c31(1)
c1(1) c1(4), c13(5), c14(7), c24(1) c24(1), c27(1)
c1(2), c6(2), c8(3), c9(4), c11(2) c1(29), c4(1), c5(4), c10(4), c11(1)
c1(10), c3(1), c11(11), c12(2)
c1(4), c7(18), c12(9), c22(1)
c8(7), c9(4), c22(10)
c1(4), c26(1) c1(3), c11(1), c13(2) c1(1), c8(1), c9(3) c22(2), c28(2)
c1(20), c2(1), c9(4), c21(2), c22(1), c25(2)
cox3 haplotypes
3
3 1 24
34
10
11
7
6
1 16 1 185
40
12
22
32
21
5 5 6 2
22
n atp6
a25(3) a26(1) a1(18), a16(1), a18(1), a19(3), a20(1) a17(3)
a30(9), a34(1)
a30(6), a33(5)
a28(1), a29(4), a30(1), a32(1)
a30(4), a31(1), a35(1)
a1(1) a1(4), a25(3), a27(6), a36(3) a36(1)
a1(3), a9(3), a11(2), a13(2), a22(2) a1(29), a4(1), a8(1), a13(1), a14(7), a15(1)
a1(12), a5(8), a6(1), a7(1)
a9(3), a11(1), a12(5), a13(1), a30(1), a36(6), a40(4) a1(11), a3(2), a21(1), a23(15), a24(1), a36(1)
a1(4), a36(1) a2(3), a25(2) a9(3), a11(1), a12(1), a40(1) a30(1), a40(1)
a1(11), a10(2), a25(1), a36(6), a39(2)
atp6 haplotypes
Table 1 Collection sites and haplotypes of mt cox3 and atp6 and cp RuBisCO spacer for Colpomenia peregrina
3
3 1 24
34
10
11
7
6
1 16 1 183
39
12
22
32
21
5 5 5 2
22
n cox3+ atp6
h44(3) h46(1) h1(2), h27(13), h28(1), h29(2), h30(1), h31(3), h34(1), h35(1) h32(1), h33(2)
h56(5), h64(1), h65(4)
h56(3), h62(1), h67(1), h68(1), h69(1) h56(5), h57(4), h58(1), h59(1)
h56(3), h60(1), h61(1), h66(1)
h19(4), h22(1), h23(1), h24(5), h47(6), h55(4) h1(3), h5(11), h6(1), h9(1), h23(1), h36(13), h38(1), h47(1) h1(8), h3(9), h4(1), h5(2), h7(1), h8(1) h1(2), h3(2), h18(1), h19(3), h21(1), h23(1), h37(2) h1(22), h3(1), h10(1), h11(1), h12(4), h13(1), h15(2), h16(1), h17(1), h25(1), h26(4) h1(1) h1(4), h45(2), h46(3), h49(7) h49(1)
h1(10), h2(1), h20(2), h39(1), h40(4), h41(1), h44(1), h52(1), h54(1) h1(3), h42(1), h48(1) h1(2), h3(1), h14(3) h19(2), h23(1), h24(1), h55(1) h56(1), h62(1)
cox3+atp6 haplotypes
r1, r3
r3, r6
4 2
r7
r9
r9
r9, r10
r9
r2 r8 r1
r2
r3
r1, r2
r4, r6
r4
r5 r1 r4 r9
r9
RuBisCO spacer haplotypes
1
9
1
2
3
3
1 1 1 17
3
1
2
2
1
1 1 1 1
1
n RuBisCO spacer
J Appl Phycol (2014) 26:491–504 495
5 38
r1 1 h52(20) h52(21), h53(2) 20 23 90 338
r3 1 h49(16) 16
r4 1 h50(2) h49(1), h51(1) 2 2
r1 r1 h47(2), h49(17), h50(3) h47(4), h52(1) 31
genetic structure was detected among sites in the combined data (FST =0.772, p
